product-marketing innovation, skills, and firm productivity growth · 2020. 4. 7. · criscuolo and...
Post on 03-Feb-2021
1 Views
Preview:
TRANSCRIPT
-
Product-Marketing Innovation,Skills, and Firm ProductivityGrowth Martin Junge, Battista Severgnini and
Anders Sørensen
ADVANCING KNOWLEDGE
THE THINK TANK
-
Product-Marketing Innovation, Skills, and Firm Productivity
Growth∗
Martin Junge Battista Severgnini Anders Sørensen
Abstract
The role of product and marketing innovation for productivity growth is addressed using survey
and register data for the Danish economy. It is argued that marketing and product innovation
are complementary inputs and that innovation activities are skill-intensive. It is found that
product and marketing innovation in skill-intensive firms results in significantly faster produc-
tivity growth than in unskilled-intensive firms that introduce this combination of innovation
activities. More precisely, an increase in the share of educated workers of one percentage point,
increases productivity growth by around 0.1 percentage point in firms with product and mar-
keting innovation. In addition, it is found that firms that engage in product innovation but not
in marketing innovation or the other way around do not demonstrate a growth effect from their
innovation activities. It is also found that product and marketing innovation has an independent
role in productivity growth that cannot be attributed to organisational changes, even though
the majority of innovative firms engage in this latter innovation type.
Keywords: Product innovation, marketing innovation, organizational innovation, productivity
growth, educational composition.
JEL codes: J24, O31, M31
∗Battista Severgnini gratefully acknowledges financial support of the Danish Social Science Research Council-Ministry ofScience, Technology and Innovation. Martin Junge. Danish Business Research Academy, Copenhagen, Denmark. Email:mj@dea.nu. Battista Severgnini. Department of Economics, Copenhagen Business School, Denmark. Email: bs.eco@cbs.dk.Anders Sørensen (corresponding author), Department of Economics, Copenhagen Business School, and Centre for Economicand Business Research, Porcelænshaven 16A, 1st floor, 2000 Frederiksberg, Denmark. Email: as.eco@cbs.dk
1
Editor: Martin Junge, DEAPublisher: DEAYear and month: July 2012 - Updated version of the paper will be published in Review of Income and Wealth in 2015.
-
1 Introduction
This paper studies effects of marketing innovation on productivity growth. We do not claim that
marketing innovation necessarily play an independent role in productivity growth; rather, we hy-
pothesise that the complementarity between product innovation and marketing innovation is an
important growth driver. The motivation for this hypothesis is that the ability to come up with
new and improved products and the ability to commercialise these products into demand seem
to require coordination between the two activities. Empirical support for product and marketing
innovation to play an independent role in productivity growth is established. More precisely, the
combination of the the two innovation activities result in significantly faster productivity growth
in skill-intensive firms.
Marketing innovations involve the implementation of new marketing practices. Specifically,
these include changes in sales and distribution methods, changes in product design and packaging,
and methods for pricing goods and services. This concept of marketing innovation follows the
definition in the Oslo Manual, OECD (2005); a concept applied in the Community Innovation
Survey (CIS) from which the innovation data used in this paper stem.1
During the past decade, the recognition that non-technological growth drivers are important for
productivity growth has emerged. One prominent driver is organisational change. Caroli and Van
Reenen (2001) argue that such changes have an independent role in productivity growth. Moreover,
the authors find that organisational changes are skill-intensive, which implies that the growth effect
of the activity is larger in skill-intensive firm than in unskilled-intensive firms. In another line of
research, Bloom and Van Reenen (2007) establish that good management practices are important for
firm productivity. More precisely, the authors find that firms with "better" management practices
tend to be more productive than firms with "worse" practices and that management practices vary
tremendously across firms, which suggests that large improvements in productivity are possible.2
For a survey on the empirics of organisational changes see Bloom, Sadun, and Van Reenen (2010).
Complementarities between non-technological and technological growth drivers are also impor-
tant for firm performance. The relationship between organisational changes and new technologies
has been shown to be important for productivity in micro studies, see for example Brynjolfsson
1 In the applied version of the Danish CIS, CIS 4, no questions have been asked related to new methods of pricinggoods and services. Questions related to this aspect have been introduced in later versions of the survey.
2Growth drivers related to technological development are also important; see Hall et al (2010) on the return toR&D and Hall (2011) on product and process innovation and productivity. The latter paper finds that there aresubstantial positive impacts of product innovation on productivity levels but that the impact of process innovationis more ambiguous.
2
-
and Hitt (2000) for a survey. The evidence shows that a combination of investment in informa-
tion technology (IT) and changes in organisations contributes to firm productivity growth. Crespi,
Criscuolo and Haskel (2007) use data for the UK and find a positive effect on firm performance
of the interaction between IT investment and organizational innovation. Moreover, Bloom, Sadun,
and Van Reenen (2012) find that the IT-related productivity advantage in the US is mainly related
to "tougher" human resource management in US-owned firms compared to European owned firms.
The contribution of this paper to the literature is to investigate another complementarity be-
tween non-technological and technological growth drivers; namely, the complementarity between
marketing innovation and product innovation. The importance of marketing — a non-technological
growth driver — has rarely been studied in relation to firm performance.3 In the marketing liter-
ature, the idea that the success of new and better products is related to marketing activities is
suggested by Gupta, Raj, and Wilemon (1986). It is argued that the success of research and de-
velopment (R&D) is potentially influenced by R&D-marketing integration. In an empirical study,
Dutta, Narasimhan, and Rajiv (1999) find that the interaction between marketing and R&D ca-
pabilities is correlated with firm performance. The study examines a cross-section of firms and
focuses on the US semiconductor industry. Moreover, Park (2004) finds in a case study that the
succes of the VHS format for the videocassette recorder market at the expense of Betamax is partly
explained by marketing ability.
It is important to understand the distinguishing factors between product and marketing innova-
tion when studying the complementarity between the two innovation activities. The Oslo manual
states that "the main distinguishing factor for product and marketing innovations is a significant
change in the product’s functions or uses." As an example, product innovation takes place when
functional or user characteristics of existing products are significantly improved. Marketing innova-
tion takes place when the design of an existing product changes significantly but this is not product
innovation, as long as the functional or user characteristics are not changed significantly.4
In the present study, we formulate and test two hypotheses. Motivated by the above-mentioned
studies of the integration of R&D and marketing, we formulate a hypothesis regarding the com-
plementarity between product and marketing innovation. In relation to this, we must emphasise
3 In the economic literature, marketing has sometimes been studied in relation to international trade. For example,Molina and Muendler (2010) find that firms hiring workers from marketing-related occupations at former exporterspredicts a wider reach of destinations for the export decision.
4For services, "the main distinguishing factor for service innovations and marketing innovations is whether theinnovation involves a marketing method or a service (i.e. a product). The distinction may depend on the nature ofthe firm’s business."
3
-
that product innovation has a much broader definition than R&D and that we do not restrict the
analysis to a narrowly defined industry. The first hypothesis states that activities in product and
marketing innovation have a positive effect on productivity growth and that the effect is larger
than for activities in either product og marketing innovation. The motivation for this hypothesis
is that product innovation generates new and improved products that potentially shift the demand
curve of the firms outwards. However, to effectively approach existing or new markets with new
or improved products, the firms will use marketing tools. In this respect, firms may indeed use
marketing innovation to approach and reach their markets. In sum, marketing innovation is ex-
pected to be an important activity for product innovation to generate growth effects. The second
hypothesis states that innovation is a skill-intensive activity.
Combining the two hypotheses, we study whether the introduction of product and market-
ing innovation in skill-intensive firms results in significantly faster productivity growth than in
unskilled-intensive firms that introduce these types of innovation. This joint hypothesis cannot be
rejected, since an increase in the share of educated workers of one percentage point, significantly
increases productivity growth by around 0.1 percentage points in firms with product and marketing
innovation. In addition, we find that firms that engage in product innovation but not marketing
innovation or firms that engage in marketing innovation but not product innovation do not show a
positive growth effect from these innovation activities.
Although the majority of innovative firms also engage in organisational innovation, we are
able to rule out the concern that the estimated effect of product and marketing innovation is
driven by organisational changes. More precisely, we find that product-marketing innovation as
well as organisational innovation play independent roles for productivity growth. These results
are obtained using two approaches. Under the first approach, we disentangle the growth effects
for different innovation types. Due to multicollinearity problems, it is not possible to disentangle
the separate effects of product and marketing innovation. As a result, we turn to an alternative
approach under which we estimate growth effects for different firm types categorised after their
innovation activities.
As a final remark, studies on innovation and firm productivity usually relate product innovation
to productivity levels; see Hall (2011). Thus, the studies mainly apply cross-sectional data, and
the results are of a long-run nature and do not analyse any dynamic responses. The present study
focuses on productivity growth - not productivity levels - and thereby focuses on the dynamic
effects of product innovation.5
5Mairesse and Mohnen (2010) survey research based on Community Innovation Surveys. In their survey they
4
-
The paper proceeds as follows. The next section proposes a simple model of production and de-
mand used to analyse innovation and firm performance. Section 3 describes the data and definitions
used. Section 4 presents the empirical results. Section 5 concludes the paper. The relationship
between the two approaches used in the empirical section are discussed in an appendix.
2 Model
2.1 Production and demand
The model we have in mind follows the model presented in, e.g., Hall (2011). It consists of the
following components: knowledge production functions, a production function and a demand func-
tion.
Firms undertake innovation activities in order to gain a monopolistic advantage or a more
effective production process. For simplicity, we assume that firms face a constant elasticity demand
function of the following form:
ln = 0 ln + 1 ln
with 0 0 and 1 0. denotes the demanded quantity, denotes the (relative) output price of
the firm, and denotes the knowledge capital stock that is relevant for demand. constitutes
a measure of the monopolistic advantage of the firm. Moreover, firms produce according to the
following production function:
ln = 0 + 1 ln + 2 ln+ 3 ln
where denotes the physical capital stock, denotes employment, and denotes the knowledge
capital stock relevant for production; a measure of the effectiveness of production.
Combining the demand and the production function results in the following function for real
refer to studies that investigate complementarities between innovation types based on cross-sectional data. Somestudies focus on complementarities between product and process innovation, process and organizational innovation,and product and organizational innovation. Junge, Severgnini, and Sørensen (2012) study the long-run nature ofthe impact of innovation types on productivity levels using a two stage model. In this paper, evidence for comple-mentarities between product and marketing innovation as well as between process and organizational innovation isestablished.
5
-
value added:
ln = ln + ln (1)
=0 + 1
0(0 + 1 ln + 2 ln) +
3 (0 + 1)
0ln − 1
0ln
where denotes real value added. In the next sub-section, we turn to the knowledge function of
the model.
2.2 Knowledge Creation
Firms can conduct product innovation and/or marketing innovation. In addition to these innova-
tion types, firms can also engage in organisational innovation. We include the latter innovation
type because it is an important growth driver, as discussed in the introduction, and because the
majority of firms that conduct product innovation also conduct organisational innovation, as de-
scriped below. The three innovation types are denoted by (product innovation), (marketing
innovation), and (organisational innovation), respectively.6 Innovation activities potentially af-
fect real value added through the growth rate of knowledge capital, i.e., ∆ ln and ∆ ln. We
assume that knowledge production relevant for demand, , is affected through product innova-
tion and marketing innovation, whereas organizational innovation is important for the knowledge
production relevant for the production function, .
An important element that we study is the complementarity between product and marketing
innovation in the knowledge creation of . We hypothesise that marketing innovation strengthens
the growth effect of product innovation (and the other way around). In other words, firms that
expend the effort to bring their products to market by developing sales and distributions methods
or by changing the design or packaging in addition to innovating new and improved products are
expected to perform better than firms that do not. Another important aspect is that innovation is
assumed to be a skill-intensive activity. The introduction of innovation activities in skill-intensive
firms is expected to lead to significantly faster productivity growth than the introduction of inno-
vation activities in unskilled-intensive firms.
The innovation process generates new knowledge and the development is assumed to follow the
6Firms can also perform process innovation. This innovation type is not included in the analysis because it isout of the scope for the analysis. Even if this type was included no empirical importance is found. The results areavailable upon request.
6
-
equations
∆ ln = ¡
¢ (2)
∆ ln =
where , , and denote intensities of product, marketing and organizational innovation,
respectively. denotes the education mix, where high values indicate skill-intensive firms and low
values indicate unskilled-intensive firms. Therefore, a key explanatory variables in the knowledge
production function is the education mix of the firms. We do not target specific departments
of firms when focusing on the relationship between the education mix and knowledge production
(i.e., we do not restrict the analysis to employees in R&D or other specific departments of firms).
Rather, we assume that a more intensive overall use of educated workers increases the intensity of
the different innovation types and that these activities can be performed in any part of a firm.7
Furthermore, it is assumed that product and marketing innovation are complementary, which
implies that: á
¢
! =
á
¢
! =
2¡
¢
0
The effect of more intensive product innovation activities increases the effect of marketing innovation
on knowledge creation (and the other way around).
Moreover, the assumption of innovation as a skill-intensive activity implies that
2∆ ln
0
2∆ ln
0
2∆ ln
= 1 0
7Variation in marketing capabilities across firms may also be important for the growth effect of production in-novation, in the sense that a high marketing capability may lead to a larger growth effect than for low marketingcapability. We do not study this aspect because we do not have data on marketing capabilities.
7
-
2.3 Real Value Added Growth
Using (1) and (2), we derive the growth rate of real value added as a function of innovation activities:
∆ ln =0 + 1
0(∆ + 1∆ ln + 2∆ ln)
+3 (0 + 1)
0 − 1
0
¡1
+ 2 + 3
¢ +
where we treat 0 = + + , with ∆ = . Moreover, we model function () as
¡
¢= 1
+ 2 + 3
. In sum the model to be estimated in the empirical section
below equals:
∆ ln = 0 + 1∆ ln + 2∆ ln+ 1 + 2
+ 3 + 4
+ (3)
where the coefficients are determined by 0 = (0 + 1)∆0, 1 = (0 + 1)10 etc. Our
main interest is in the interaction term, i.e., 3 = −130. The advantage of estimating theregression in first differences is that it includes firm-specific fixed effects to control for unobserved
firm heterogeneity.
3 Data and definitions
The applied data set stems from two sources. First, data on innovation activities originate from the
Danish version of the Community Innovation Survey (CIS) from 2004 that constitutes a representa-
tive sample for the Danish economy. This data set includes information on innovation activities at
the firm level. The questions on innovation activities are related to innovation activities performed
in firms during the 3-year period from 2002-2004. Second, other firm variables that originate from
the Danish register data sets IDA and FIDA are included. Specifically, the backbone of the analy-
sis is an employer-employee matched data set that enables us to track educational attainment for
each employee in every single firm. Consequently, it is possible to obtain very precise measures of
the education mix applied in each firm. Moreover, firm information in the data set provides data
on value added, labour input, and input of physical capital. In addition, we also apply industry,
regional, and ownership dummies as well as export intensities. Finally, the register data allows the
possibility of measuring the variables for different years. We apply and describe this possibility
below.
8
-
3.1 Innovation
In the empirical analysis three types of innovations are applied. The three types are product
innovation ( ), marketing innovation (), and organisational innovation ():
A firm is considered to perform:
• product innovation if it answers yes to the question: ”Did the firm introduce new or signifi-cantly improved products or services?”
• marketing innovation if it answers yes to one or both of the questions: ”Did the firm introduce(i) new or significantly changed sales or distribution methods or (ii) significant changes in a
product or service design or packaging?”
• organisational innovation if it answers yes to one or both of the following questions: ”Did thefirm introduce (i) new or significantly improved business processes for better use or exchange
of information, knowledge and skills in the firm or (ii) a major organisational change within
the company?”
Unfortunately, we cannot observe innovation intensities. Based on the innovation questions,
we therefore construct binary indicators from the survey data that indicate whether firms had
innovation activities of the three types during 2002-2004. These binary variables are used as
measures for innovation activities below.
3.2 Other firm-specific variables
Real value added, , is measured as value added in current prices deflated by narrow industry
specific price series. Physical capital, , is measured as fixed assets and stems from accounting
data, whereas labour input, , is measured by full time equivalent units. These variables are used
to determine long log-differences between 2004 and 2007. These log differences are measured as
annualised changes. As a measure of the education mix in the single firm, , we take the share of
full-time employees in the firm with at least 16 years of education out of total number of employees.
By this measure, the higher is the more skill-intensive is a firm. The descriptive statistics of the
sample are shown in Appendix Table 1.
9
-
4 Empirical results
This section presents the empirical results of the study. First, we comment on a number of econo-
metric problems that are dealt with in the analysis. Second, we present the descriptive statistics,
and finally, we present the results of the regression analysis.
4.1 Econometric Problems
There are a number of econometric problems in estimating the relation between productivity and
the knowledge stock of the firm. Specifically, these problems are unobserved heterogeneity and
endogeneity. We deal with unobserved heterogeneity by estimating the model in first differences,
as discussed in relation to (3) above. Consequently, we apply annualised changes during the period
2005-2007. To mitigate endogeneity, we use lagged innovation variables in the regressions, see
Griffith, Redding, and Van Reenen (2004). More precisely, we use data for innovation activity for
the period 2002-2004. Moreover, we apply an initial skill share; the education share measured in
the first week of November 2001, i.e., the share is measured before the period that the innovation
measures apply for. The timing of the dependent and explanatory variables is illustrated in Figure
1.
Figure 1 around here
To deal with a potential selection problem, we include lagged growth rates in one of the regres-
sions in the robustness section, i.e., Section 4 below. That is we include the annualized log-change
in real value added over the period 2002-2004 as explanatory variable. The issue is that high-
growth firms potentially select themselves into innovation activities and low-growth firms do not.
More precisely, the lagged growth rates are instrumented with the initial real value added level as
measured in 2001, which follows Caroli and Van Reenen (2001).
4.2 Descriptive Statistics
Innovation activities for the firms in our sample are presented in Table 1. In Table 1a, the firms
are grouped after whether they perform product and/or marketing innovation, whereas they are
grouped after product, marketing and organisational innovation in Table 1b.
Table 1 around here
10
-
In Table 1a, it is evident that 527 firms out of 1,321 firms - representing 40 per cent of the firms
in the sample - engage in product innovation. Of these firms, approximately one third conduct
marketing innovation in addition to product innovation, whereas the remaining two thirds do not
conduct marketing innovation. Moreover, it is seen that 261 firms engage in marketing innovation.
Of these firms, about 60 per cent also perform product innovation.
In Table 1b it is seen that 401 firms out of 527 firms conducting product innovation also engage
in organisational innovation; i.e., three quarters of the firms conducting product innovation also
perform organisational innovation. For the group of firms engaging in both product and marketing
innovation, as many as 83 per cent also conduct organisational innovation. It is also evident
that 307 firms - nearly one fourth of the sample - engage in organisational innovation without
conducting either marketing or product innovation. Approximately 60 per cent of the firms engage
in organisational innovation, implying that the majority of the 71 per cent of firms in the sample
that conduct innovation activities are conducting organisational organisation. 29 per cent of firms
do not engage in innovation activities.
Before turning to the regression analysis, we present the average measures of labour productivity
growth divided by innovation type and skill intensity. In Figure 2, the average growth rates of skill-
intensive firms are compared to the average growth rate of all firms. The growth rates are presented
as deviations from the average growth rate for different firm types defined after the innovation
activities that they engage in. Moreover, firms are considered to be skill-intensive if their education
share in 2001 exceeds the median share within their firm type.
We split the data into four types of innovative firms according to the innovation activities that
they engage in. The firms are grouped according to whether they conduct (1) product-marketing
innovation (), (2) product innovation but no marketing innovation (), (3) marketing
innovation but not product innovation (), or (4) organisational innovation but no product
or marketing innovation (). Finally, we also include firms without innovation activities
().8
Figure 2 around here
8We only apply 4 firm types with innovation activities in the main text to reduce the number of parametersto estimate. This is the case even though the combinations of three innovation types result in 7 firm types withinnovation activities. See the appendix for a discussion.
11
-
A number of interesting observations are shown in Figure 2. It is clear that skill-intensive firms
with innovation activities grow faster than the average firm within the same firm types and thereby
unskilled-intensive firms. This effect is found for all firm types with innovation and suggests that
innovation is indeed a skill-intensive activity. It is especially pronounced for firms that engage in
product and marketing innovation. A second and somewhat surprising observation is that within the
group of firms without innovation activities, unskilled-intensive firms grow faster than skill-intensive
firms. This observation should not be taken at face value, but rather understood as indicating that
skill-intensive non-innovators do not grow faster than unskilled-intensive non-innovators. In other
words, this observation suggests that growth effects of skills to a large extent work through the
innovation channel, i.e., skills are contributing to knowledge creation.
There are also a number of things we cannot determine from Figure 2. We cannot determine
whether the growth effects are significantly different from zero. We do not control for other back-
ground characteristics of the firm than the skill intensity. Moreover, it is not possible to contribute
growth effect to the different innovation activities. Most problematic is the growth performance of
product innovation and/or marketing innovation because the majority of innovative firms perform
organizational innovation. This is especially outspoken for firms that perform both product and
marketing innovation, which was evident from Table 1. In this sense, we cannot determine whether
it is organizational innovation, product-marketing innovation or both that drives the growth per-
formance in firms. Consequently, we have to disentangle the effect of organisational innovation
from the other types of innovation. These concerns will be taken into account in the next section.
4.3 Regression results
In the following two sets of regression results are presented. First, we focus on the growth effects
of different innovation types. That is, we attempt to disentangle the growth effects of the three
innovation types: product, marketing, and organisational innovation. From an empirical perspec-
tive, disentangling effects of the different innovation types is challenging because most of these
coexist in the production function and thus create possible problems of collinearities (Anderson
and Schmittlein (1984), Milgrom and Roberts (1990), and Athey and Stern (1998)). As it turns
out, we are able to obtain precise estimates of the combined growth effects of product and market-
ing innovation as well as organisational innovation. It is, however, not possible to determine the
separate effects of product innovation or marketing innovation. As a consequence we focus on firm
types that are engaging in different types of innovation rather than focusing on different innovation
types. More precisely, we present results for the four types of innovative firms defined in relation
to Figure 2 above. The two approaches are derived from the same general model. Moreover, the
12
-
empirical results obtained under the two approaches are qualitatively the same. These aspects are
discussed in Appendix A.
4.3.1 Innovation types
First, we present the results for the growth effects of different combinations of product and market-
ing innovation in Table 2. Next, we include organisational innovation in addition to product and
marketing innovation in Table 3.9
In Table 2, the regression results of including product innovation, marketing innovation, and
the interaction term of the two innovation types are presented. In the first two columns, we
introduce the two innovation types one at a time. In column 1, product innovation is introduced
and the relationship between growth in real value added and this innovation type is estimated. The
interaction term between product innovation and the education share, i.e., the coefficient of ·,is positive and significantly different from zero. The point estimate for the interaction term equals
0.087 and is significantly different from zero at the 5 per cent level. This implies that skill-intensive
firms that engage in product innovation grow faster than skill-intensive firms that do not engage
in product innovation. More precisely, an increase in the skill intensity of one percentage point,
increases productivity growth by almost 0.1 percentage points in firms with product innovation.
Table 2 around here
It is also evident in column 1 that the direct coefficient to product innovation, i.e., the coefficient
of , is approximately zero, implying that the average firm does not obtain any growth effect for
product innovation. This also means that the skill intensity may be too low to generate positive
growth effects, since the demeaned effect of · is negative for firms with a skill share belowaverage. I.e., the growth effect may actually be negative, if the skill intensity is below a threshold
level. Taken at face value, firms with a skill intensity below 0.1663 will experience a negative
growth effect of conducting product innovation, whereas firms with a skill intensity above 0.1663
will have positive growth effects. This difference suggests that innovation only pays off growth-wise
in skill-intensive firms. This effect is also found for organisational innovation in Caroli and Van
Reenen (2001).
9 In the regression analysis, the interaction terms are determined as the interaction between innovation dummiesand the demeaned skill-intensity. Subtracting the mean of the skill share, ensures that average treatment effect (ATE)is the coefficient on the innovation dummy itself.
13
-
In column 2, marketing innovation is included as the only innovation type in the growth regres-
sion. As for product innovation, the interaction term between innovation and the skill intensity
enters with a positive value; however, in this case, the value is not significantly different from zero.
Column 3 presents the results of including both product and marketing innovation as separate and
non-complementary activities. The findings in column 3 reflect those of columns 1 and 2.
Column 4 includes the interaction between product and marketing innovation. It is clear that
the estimates are influenced by multicollinearity problems. All three estimates for the innovation
types are imprecise. The combined growth effect of conducting product and market innovation
simultaneously when the two types of innovation are treated as complementary is equal to 0.131
(=0.045-0.120+0.206); this value is significant at the 1 per cent level when using an F-test, which
is shown in the bottom part of the column.
Because the direct effect of marketing innovation was found to be insignificant in column 2,
it is excluded in column 5. By excluding this effect, we find positive but insignificant separate
effects of product innovation. Moreover, the combined effect of engaging in product and marketing
innovation equals 0.141 (=0.059+0.086), which is of similar magnitude to the total effect found
in column 4. Again, this effect is found to be significant at the 1 per cent level using an F-test.
Finally, the direct effect of product innovation is also excluded in column 6. In this case, we find
that the coefficient for the interaction between product and marketing innovation is equal to 0.127,
which is significant at the 1 per cent level.
The results of Table 2, however, can be questioned: is the growth effect of product and marketing
innovation actually captured by the estimated coefficients in the growth regressions? As was evident
from Table 1, three quarters of firms with product innovation and/or marketing innovation also
engage in organisational innovation. Therefore, the growth effect from organisational innovation
may be picked up by the applied measures of product and marketing innovation. Another important
reason to worry about the growth driving effect is articulated by Caroli and Van Reenen (2001), who
study growth effects of organisational changes. This work argues that organisational change has
an independent role in productivity growth. Moreover, the results demonstrate that organisational
changes are skill-intensive, as was found for product-marketing innovation above.
For these reasons, we include organisational innovation in the growth regressions in addition to
product and marketing innovation. The results are presented in Table 3. In column 1, organisational
innovation is included as the only innovation type. It is shown that organisational innovation
activities have a positive growth effect for skill-intensive firms, a result that parallels the findings
14
-
in Caroli and Van Reenen (2001). In this sense, organisational innovation indeed has an important
role to play for economic growth.
Table 3 around here
In column 2, we include product and marketing innovation in addition to organisational innova-
tion. Again, the estimation for product and marketing innovation suffers from multicollinearity due
to the innovation structure, as presented in Table 1. Therefore, we evaluate the effect of conducting
product-marketing innovation as the total effect of product innovation and marketing innovation.
The combined effect is positive and equal to 0.099 (=0.196-0.120+0.022), which is significant at the
10 per cent level when using an F-test. In other words, we find lower significance when including
organisational innovation but the effect is still positive and significant. As in column 1, we find that
organisational innovation has a positive growth effect in skill-intensive firms; however, the effect is
no longer significant. This result may well be due to the fundamental multicollinearity problem.
In columns 3 and 4, the direct effects of marketing and product innovation are removed one at
a time, as was the case in Table 2. In column 3, marketing innovation is removed. In this case
the common effect for product and marketing innovation is equal to 0.113 (=0.077+0.036) and is
significant at the 5 percent level. When both direct effects are removed, the coefficient for the
interaction term between product and marketing innovation equals 0.100 that is also significant
at the 5 percent level. The coefficients for organisational innovation, · , are positive in bothspecifications; however, the coefficient of column 3 is insignificant, whereas it is significant at the
10 percent level in column 4.
In sum, we find that skill-intensive firms that both perform product and marketing innovation
experience a positive growth effect. Since we are able to disentangle the effects of product-marketing
innovation and organisational innovation, we can conclude that this effect is not driven by organisa-
tional innovation. An increase in the share of educated workers of one percentage point, increases
productivity growth by around 0.1 percentage points in firms with product and marketing in-
novation, whereas productivity growth increases by around 0.08 percentage points in firms that
perform organisational innovation. The positive growth effect may require that the skill intensity,
as measured by the share of employees with 16 or more years of education, exceeds a threshold
level.
As final remarks it should be mentioned that the skill-intensity itself does not contribute to
growth, i.e., skills only have a growth effect through innovation. This is the case for all regressions
15
-
reported in Tables 2 and 3 since the point estimates on the lagged skill-intensity is insignificantly
different from zero in all the regressions; a result that was also evident in Figure 2 above. Moreover,
the conventional factors labor and capital take their expected signs. The point estimate on growth
of labor equals 0.84, which is close to the share of labor costs in value added and it is in line
with estimates found elsewhere in the literature. The point estimate on growth of capital equals
0.03. Even though it is a relatively low value this magnitude is found elsewhere, see for example
Balsvik (2011). One potential explanation for the low point estimate is that capital is measured
by fixed assets that origin from accounting data; a measure that does not measure physical capital
as precise as we would prefer. This implies that capital has measurement errors that is likely to be
exacerbated as a consequence of differencing.
Due to the low point estimate on growth of capital the hypothesis of constant returns to scale in
labor and capital is rejected at the 1 per cent level when using a F-test. In spite of this inappropriate
result, the main conclusions derived above and in the next section are unaffected by the inclusion of
poorly measured physical capital. We conclude this because the qualitative results for the growth
effects of innovation is unaffected when constant returns to scale is enforced on the regressions.
4.3.2 Firm types
An alternative estimation approach that deals with the multicollinearity problem of different inno-
vation types is to estimate growth effects of different firm types. Under this alternative approach,
we split the firms into the 4 types of innovative firms used in Figure 2.
The results are presented in Table 4. It is shown in column 1 that skill-intensive firms that
conduct product-marketing innovation have higher growth effects of innovation than firms that
conduct either product innovation or marketing innovation. Firms that engage in product innova-
tion but not marketing innovation experience a positive growth effect, however, the effect is not
significantly different from zero. Moreover, the growth effect for firms with marketing innovation
but no product innovation is negative but insignificant.
Table 4 around here
Again, it should be emphasised that firms belonging to one of the three firm types may also
conduct organisational innovation in addition to product and/or marketing innovation. Therefore, it
is difficult to determine whether the growth effect stems from product and/or marketing innovations
or whether it is due to organisational innovation. Therefore, the strategy applied in the following is
16
-
to test whether skill-intensive firms with product and/or marketing innovation have a significantly
higher growth effect than firms that have organisational innovation only.
In column 2, we also include firms that engage in organisational innovation but no marketing
innovation or product innovation. In this case, we establish that firms with product-marketing
innovation have higher growth effect than firms with organisational innovation only. To investigate
whether this difference is significantly different from zero, we perform a one-sided Wald test. The
difference is found to be significant, which is clear from the F-test in the bottom part of column
2. This implies that the null hypothesis (no or a negative difference) is rejected with a p-value of
0.056. We interpret the positive difference between the two coefficients as the contribution from
product-marketing innovation. This implies that an increase in the share of educated workers of
one percentage point, increases productivity growth by 0.086 (=0.168-0.082) percentage points in
firms with product and marketing innovation.
Firms with product innovation but no marketing innovation and firms with organisational in-
novation only demonstrate growth effects from innovation of similar magnitudes. This similarity
suggests that product innovation in itself does not contribute to economic growth because a large
fraction of the firms conducting product innovation also conduct organisational innovation. Firms
with marketing innovation but no product innovation has a negative but insignificant growth effect
from innovation. Consequently, product and marketing innovation in itself do not contribute to
productivity growth.
To come closer to the additional effect of product-marketing innovation over and above the
effect of organisational innovation, we exclude innovative firms that do not engage in organisational
innovation. This means, for example, that firms that conduct product-marketing innovation but
no organisational innovation are excluded from the sample in column 3. However, firms without
innovation activities are included. Using this restriction, 163 innovative firms are excluded, resulting
in a sample of 1,158 firms. The results for the restricted sample imply that skill-intensive firms that
engage in product, marketing, and organisational innovation have significantly higher growth effects
than skill-intensive firms that conduct organisational innovation only at the 10 percent significance
level. The impact on productivity growth of an increase in the skill share of one percentage point
equals 0.070 (=0.155-0.085) percentage points.
Firms with product and organisational innovation but without marketing innovation and firms
with marketing and organisational innovation but without product innovation have a larger growth
effect than firms with organisational innovation only. The differences are, however, not statistically
17
-
significant implying that product innovation without marketing innovation and marketing inno-
vation without product innovation do not contribute significantly to productivity growth. It is
interesting that the point estimate to marketing innovation becomes positive when excluding firms
performing marketing innovation only instead of a negative effect when included as in columns 1
and 2.
Finally, in column 4 we only include firms that conduct organisational innovation, implying that
we exclude firms with innovation activities that do not perform organisational innovation, from the
sample. This exclusion results in a sample of 776 innovative firms presented in Table 1b. We find
that product-marketing innovation generates higher growth effects in skill-intensive firms than in
firms that conduct product innovation but not marketing innovation, marketing innovation but
not product innovation or organisational innovation only. Only the interaction term for firms with
product-marketing innovation is significantly different from 0.
In summary, the conclusion drawn from Table 4 is that skill-intensive firms with product-
marketing innovation experience a positive growth effect in the 3-year period after the innovation
activity took place. The growth effect in skill-intensive firms engaging in product-marketing in-
novation significantly exceeds that of firms conducting organisational innovation only. This is an
important result because many firms that conduct product-marketing innovation also conduct or-
ganisational innovation. To rule out the possibility that the growth effect for product-marketing
innovation is driven by organisational innovation, the growth effect for firms performing product
and marketing innovation must exceed that of firms engaging in organisational innovation only.
This is exactly what we find. Moreover, we find that skill-intensive firms that engage in product
and organisational innovation do not have a higher growth effect than skill-intensive firms that
engage in organisational innovation only. This suggests that product innovation without marketing
innovation does not contribute to productivity growth. A similar result is found for marketing
innovation.
Taking the results in Tables 2-4 as a whole, the evidence is clear that product-marketing inno-
vation and organisational innovation both generates positive growth effects and that innovation is
a skill-biased activity.
4.3.3 Robustness
Finally, we investigate the robustness of the results established for product and marketing innova-
tion in the section above. We present a number of alternative specifications in Table 5. In column
18
-
1 of the table, we present our baseline specification of the previous sections, namely column 2 of
Table 4. We compare this specification against the alternative regressions.10
Table 5 around here
The first issue addressed in Table 5 is firm size. The applied binary measure of product inno-
vation may be problematic because large firms have a tendency to answer "yes" to the question of
whether they engage in product innovation or not. In the words of Hall (2011), p. 176, "When
examined across a range of firm sizes, it produces the misleading results that larger firms are more
likely to be innovative, whereas in truth larger firms are involved in a wider range of activities and
are therefore more likely to have an innovation in at least one of them."
We expect the "size" problem to be especially pronounced for product innovation. We do not
expect firm size to be an important problem for marketing and organisational innovation. Market-
ing innovation will to a large extent be focussed on overall firm demand, whereas organisational
innovations are expected to be implemented for the firm as a whole.
The results when splitting the sample by firm size are presented in columns 2 and 3. Column
2 presents firms with employment below the median level of 60 full time equivalent workers. It is
found that product-marketing innovation has a high growth effect for skill-intensive firms below
the median size. Organisational innovation also has a positive growth effect; however, this effect is
not significantly different from zero. As in the baseline case, marketing innovation has a negative
but insignificant growth effect. The flop side result, however, is found for larger firms in column
3. Here, we can see that organisational innovation constitutes an important growth driver for
larger firms, whereas firms conducting product-marketing innovation do not gain as much from
the activity. The latter finding indicates that the critique of the measure of product innovation
may be valid. Interestingly, marketing innovation without product innovation has a positive and
significant growth effect for larger firms. These results suggest that marketing and organisational
changes have positive growth effects in larger organisations.
Next, we present an economy-wide estimate of the growth effect of innovation. To do this
we use a weighted regression of the main specification, which is presented in column 4. The
estimates are weighted by establishment size and sample frequency from Statistics Denmark. The
10Similar results to those presented in this section is established for Table 3, column 4. The results are availableupon request.
19
-
regression shows that the growth effect in skill-intensive firms that conduct product-marketing
innovation is positive and significant. This is the case even though large firms that tend to answer
"yes" to engaging in product innovation as discussed above have larger weights. Skill-intensive
firms conducting organisational innovation only also have a positive growth effect. Surprisingly,
however, this effect is not significantly different from zero. When specifying the regression with
organisational innovation as the only type of information included in the regression (not shown),
we find that skill-intensive firms gain significantly from organisational innovation. Finally, skill-
intensive firms conducting marketing innovation but not product innovation have a positive but
insignificant growth effect.
A concern regarding the results established above is that they may simply be a result of the
selection of high growth firms into innovation. It may be the case that product-marketing innovation
is introduced in firms that have higher productivity growth rates (see for example Caroli and Van
Reenen (2001) and Ichniowski, Shaw, and Prennushi (1997)).
To deal with this potential selection problem, we include the lagged growth rates of real value
added in the regressions as a robustness check. In other words, we include the annualised log
change for the period 2002-2004. The issue is that high-growth firms potentially select themselves
into innovation activities and low-growth firms do not. The lagged growth rates are instrumented
with the initial real value added level measured in 2001, which follows Caroli and Van Reenen
(2001). The main result established in this paper is unaffected by this specification.11
In this section, we have performed a number of robustness checks of the baseline specification
of Table 4. Neither of these specifications change the main result that skill-intensive firms that
perform product and marketing innovation experience a positive growth effect in the subsequent
(long) period. We document that the combined innovation activity has an independent role in
productivity growth and are able to disentangle the growth driver from organisational changes. On
the contrary, product innovation or marketing innovation alone do not contribute to growth.
11Another check is to estimate the relationship between productivity growth and innovation activities using quantileregressions. The results have been estimated for the 0.50 quantile (the median), the 0.75 quantile, and the 0.25quantile and can be obtained upon request to the authors. Similar results are found for all three regressions: Thecoefficients on product-marketing innovation are of similar magnitude independent of quantile. Moreover, it is foundthat the coefficient to product-marketing innovation is significantly larger than the coefficient to firms that performorganizational innovation only for the median firm as well as for the 0.75 quantile firm. This result is not establishedwhen estimating for the 0.25 quantile.
20
-
5 Discussion and conclusion
The main result of this paper is that skill-intensive firms that engage in product and marketing
innovation grow faster than skill-intensive firms that do not engage in these activities. Moreover,
we find that firms that conduct product innovation but no marketing innovation or marketing
innovation but no product innovation do not reap higher productivity growth.
An independent issue is to separate effects of product and marketing innovation from organi-
sational innovation. In the applied data set, it turns out that the majority of firms that conduct
product and marketing innovation also conduct organisational innovation. Therefore, we study
whether firms that conduct product and/or marketing innovation gain positive growth effects in
addition to those obtained through organisational innovation. Although the majority of innovative
firms also engage in organisational innovation, we are able to rule out the concern that the estimated
effect of product and marketing innovation is simply picking up growth effects generated through
organisational innovation. Instead we find that product-marketing innovation and organisational
innovation both play independent roles in productivity growth.
One important result of this analysis is that innovation activities generate higher productivity
growth rates in skill-intensive firms, whereas they do not in unskilled-intensive firms. Indeed,
innovation activities may result in negative growth rates in unskilled-intensive firms. Taking the
estimated coefficients at face value, firms engaging in product and marketing innovation should
have an education mix of employees with 16 or more years of education in relation to the total
number of employees that exceeds the average value of the sample. A similar result is found for
organisational innovation.
The broader implication of the positive growth effect in skill-intensive firms is twofold. The
analysis finds that it may not always be appropriate to engage in innovation activities in unskilled-
intensive firms as the activity may not pay off.12 An important question in this respect is whether
firms become more skill-intensive as they begin to engage in innovation, or whether the skill intensity
to a high degree is fixed. If the latter is the case, it may be important to target skill-intensive firms
with innovation-promoting policy programs.13 In this sense, the result of the analysis may have
12This idea has been suggested for the macro-perspective in Sørensen (1999). Here, the applied endogenous growthmodel predicts that skills are accumulated as the only growth-generating when human capital is below a thresholdlevel, whereas both innovation activities and learning drive growth above this level. In steady-state equilibrium,growth is driven by innovation only.13 In the area of technological adoption a key determinant seems to be the skill-intensity of the plant. Adoption
of new technology does not typically lead to a significant increase in the firm skill-intensity (see Doms, Dunne, andTroske (1997) and Caselli and Coleman (2001)).
21
-
implications for industrial policy that is designed to increase innovation activities in firms. Instead
of targeting all firms with programs designed to increase innovation, governments may instead
direct the programs towards skill-intensive firms. Alternatively, governments should target human
capital accumulation and innovation-promoting programs simultaneously.
22
-
References
[1] Anderson, E., and D.C. Schmittlein, (1984), "Integration of the sales force: An empirical
examination". RAND Journal of Economics 15 (3), 385—395.
[2] Athey, S., and S. Stern (1998). "An empirical framework for testing theories about complimen-
tarity in organizational design". NBER Working Papers 6600, National Bureau of Economic
Research
[3] Balsvik, R. (2011), "Is Labor Mobility a Channel for Spillovers from Multinationals? Evidence
from Norwegian Manufacturing", Review of Economics and Statistics, 93(1), 285-297
[4] Bloom, N., R. Sadun and J. Van Reenen (2010), "Recent Advances in the Empirics of Orga-
nizational Economics", Annual Review of Economics, 2, 105-37
[5] Bloom, N., R. Sadun and J. Van Reenen (2012), "Americans do I.T. Better: US Multinationals
and the Productivity Miracle", American Economic Review, 102(1), 167-201
[6] Bloom, N., and J. Van Reenen (2007), "Measuring and explaining management practices across
firms and countries", The Quarterly Journal of Economics 122 (4), 1351—1408
[7] Bloom, N and J. Van Reenen (2010 ) "Why do Management Practices Differ across Firms and
Countries?" Journal of Economic Perspectives, 24(1)
[8] Bresnahan, T. F., E. Brynjolfsson, and L.M. Hitt (2002), "Information technology, workplace
organization, and the demand for skilled labor: Firm-level evidence", The Quarterly Journal
of Economics 117 (1), 339—376.
[9] Caroli. E. and J. Van Reenen (2001). "Skill-biased Organizational Change? Evidence from a
Panel of British and French Establishments", Quarterly Journal of Economics, 116(4), 1449-
1492
[10] Caselli, F., and W. J. Coleman (2001), "Cross-Country Technology Diffusion: The Case of
Computers", American Economic Review 91, 328-335
[11] Crespi, G., C. Criscuolo, and J. Haskel (2007), "Information technology, organizational change
and productivity growth: evidence from UK firms", CEP Discussion Paper no. 783.
[12] Doms, M., T. Dunne, and K. Troske (1997), "Workers, Wages, and Technology", Quarterly
Journal of Economics, 112, 253-290
23
-
[13] Dutta, S., O. Narasimhan, and S. Rajiv (1999). "Success in High-Technology Markets: Is
Marketing Capability Critical?" Marketing Science 18(4): 547-568.
[14] Gupta A. K, S.P. Raj and D. Wilemon (1986). "AModel for Studying R&D-Marketing interfase
in the product innovation process." Journal of Marketing Vol 50.
[15] Hall, B. H., J. Mairesse, and P. Mohnen (2010). Measuring the Returns to R&D. In Hall, B.
H. and N. Rosenberg, Handbook of the Economics of Innovation. Amsterdam and New York:
Elsevier.
[16] Hall, B. H., (2011). Innovation and productivity. In: Productivity and Competitiveness, eds.
J. Madsen and A. Sørensen. Nordic Economic Policy Review, 2, 167-204
[17] Griffith, R., S. Redding, and J. Van Reenen (2004). "Mapping the two Faces of R&D: Pro-
ductivity Growth in a Panel of OECD Industries." The Review of Economics and Statistics.
86(4), 883-895
[18] Ichniowski, C., K. Shaw, and G. Prennushi (1997), "The Effects of Human Resource Man-
agement Practices on Productivity: A Study of Steel Finishing Lines," American Economic
Review
[19] Junge, M., B. Severgnini, and A. Sørensen (2012), "Evidence on the Impact of Education on
Innovation and Productivity", CEBR working paper
[20] Milgrom, P., and J. Roberts (1990), "The economics of modern manufacturing, technology,
strategy and organization", American Economic Review 80, 511—528.
[21] Mairesse, J., and P. Mohnen (2010), "Using Innovation Surveys for Econometric Analysis", in
Handbook of The Economics of Innovation, B.H. Hall and N. Rosenberg (eds.), vol. 2
[22] Molina, D., and M-A Muendler (2010), "Preparing to Export", Working Paper
[23] OECD (2002). Frascati Manual: Proposed Standard Practice for Surveys On Research and
Experimental Development. OECD, Paris.
[24] OECD (2005). Oslo Manual: Guidelines for Collecting and Interpreting Innovation Data.
OECD, Paris.
[25] Park, S., (2004), "Quantitative analysis of network externalities in competing technologies:
The VCR case", The Review of Economics and Statistics 86 (4), 937—945.
24
-
[26] Sørensen, A. (1999), "R&D, Learning, and Phases of Economic Growth", Journal of Economic
Growth 4(4), 429-45, 1999
25
-
A Appendix: The general model
In Section 3.3, two approaches are applied; one based on innovation types and another on firm
types. In this appendix, we show that both of these approaches are consistent with a general
version of the model presented in (3). In the following, we concentrate on the part of the model
that is related to innovation; a part that we label :
= 1 + 2
+ 3 + 4
+ 5 + 6
+ 7
Compared to (3), we have included additional interaction terms resulting in 3 extra parameters,
5, 6, and 7.
A.1 Innovation types
The model applied under the innovation types approach, i.e., equation (3), is obtained under the
restriction that 5 = 6 = 7 = 0. This results in the following estimates ̂1, ̂2, ̂3, and ̂4:
= ̂1 + ̂2
+ ̂3 + ̂4
A.2 Firm types
Given the 3 innovation types, we can group innovative firms into seven different groups of firms
after their the innovative activities that they perform:
1 = 1 if = 1 = 1 = 1; 1 = 0 otherwise
2 = 1 if = 1 = 0 = 1; 2 = 0 otherwise
3 = 1 if = 0 = 1 = 1; 3 = 0 otherwise
4 = 1 if = 0 = 0 = 1; 4 = 0 otherwise
5 = 1 if = 1 = 1 = 0; 5 = 0 otherwise
6 = 1 if = 1 = 0 = 0; 6 = 0 otherwise
7 = 1 if = 0 = 1 = 0; 7 = 0 otherwise
Using these 7 types of innovative firms, the general version of is rewritten to:
= 11 + 22 + 33 + 44 + 55 + 66 + 77
26
-
with
1 = 1 + 2 + 3 + 4 + 5 + 6 + 7
2 = 1 + 4 + 5
3 = 2 + 4 + 6
4 = 4
5 = 1 + 2 + 3
6 = 1
7 = 2
The firm type approach is consistent with the innovation type approach when the restrictions
5 = 0, 6 = 0, and 7 = 0 are implemented. In this case, 1 = 1 + 2 + 3 + 4, 2 = 1 + 4,
3 = 2 + 4, and 4 = 4, 5 = 1 + 2 + 3, 6 = 1, and 7 = 2.
In order to avoid using 7 different firm types in the main text, the baseline regression in Section
3.3. is not based on these restriction. Instead we base the analysis reported in the main text on 4
firm types with innovation activities. The model applied under the firm types approach is obtained
under the following restrictions.
• Firm types 1 in the main text is defined as firms with product-marketing innovation. Thesefirms are identified for 1 = 1 or 5 = 1. These firm types are assumed to have similar growth
effects, implying that: 15 = 1 = 5 or 4 + 5 + 6 + 7 = 0
• Firm type 2 in the main text is defined as firms with product innovation but no marketinginnovation. These firms are identified for 2 = 1 or 6 = 1. These firm types are assumed to
have similar growth effects, implying that: 26 = 2 = 6 or 4 + 5 = 0.
• Firm type 3 in the main text is defined as firms with marketing innovation but no productinnovation. These firms are identified for 3 = 1 or 7 = 1. These firm types are assumed to
have similar growth effects, implying that: 37 = 3 = 7 or 4 + 6 = 0.
• Firm type 4 in the main text is defined as firms with organisational innovation but no productinnovation or marketing innovation. There is no restrictions for the relevant parameter 4.
Thereby, we allow firms in the first group of firms and firms in the fifth group of firms to have
similar effects on productivity growth, even though firms in the former group have organisational
27
-
innovation in addition to product-marketing innovation. This is also the case for firms with product
innovation but no marketing innovation and for firms with marketing innovation but no product
innovation. This is of course inconsistent with the applied theoretical model, since we ignore that
firms belonging to the first three groups of firms have organisational innovation in addition to
product and/or marketing innovaton.
Implementing the restrictions described in the bullits above imply that can berewritten
to
= ̂15 (1 + 5) + ̂26 (2 + 6) + ̂37 (3 + 7) + ̂44
It is clear that the two approaches lead to different estimates, i.e., ̂15 6= ̂1 etc. In other words,the two approaches origin from the same general model, but are not nested. Under the firm type
approach, the constraints imply that 4 = −5 = −6 = 7. In this sense, other complementaritiesthan the complementarity between product and marketing is enforced by the constraints. The two
approaches, however, lead to identical estimates under the restriction that 4 = 4 = 0, which is
evident from Table 2, column 4, and Table 4, column 1, that present identical results. Moreover,
the two approaches lead to identical results if all innovative firms with innovation activities perform
organisational innovation. Results for this case is reported in Table 4, column 4.
As mentioned, we do not present estimation results for the 7 groups of firms. The qualitative
results for this case, however, are similar to those presented in Table 4, implying that 5 6= 0, 6 6= 0,and 7 6= 0 do not have important implications for the results. When testing these contraints forthe firm type approach simultaneously, the F-test has a critical p-value of 0.55 implying that the
restrictions cannot be rejected. These results are available upon request.
28
-
Figure 1: Timing of variables in regressions.
Figure 2: Labor Productivity Growth, 2005‐2007. (Deviation from average growth rate in skill‐intensive firms from mean). Groups of Firms Divided after types of Innovation Activities.
Note: Firms are considered to be skill‐intensive if their education share in 2001 exceeds the median share within their firm type
‐1.0%
‐0.5%
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
PM‐firms PNM‐firms NPM‐firms ONPNM‐firms NI‐firmsLabo
r prrod
uctiv
ity growth, 200
5‐20
07
E > median E
-
Table 1A: Firms with Different Types of Innovation Product Innovation Yes No Total
Marketing Innovation
Yes 164 105 269 No 363 689 1,052 Total 527 794 1,321 Table 1B: Firms with Different Types of Innovation
Organisation Innovation
Product Innovation Yes No Total
Marketing Innovation
Yes 136 68 204 No 265 307 572 Total 401 375 776
No organisation Innovation
Product Innovation Yes No Total
Marketing Innovation
Yes 28 37 65 No 98 382 480 Total 126 419 545
-
Table 2: Firm Level Productivity Functions. Product and Marketing. Change in ln(Value added 2004-2007) (annualized mean =.024) 1 2 3 4 5 6 dlnL 0.840 *** 0.837 *** 0.840 *** 0.842 *** 0.840 *** 0.839 *** 0.035 0.035 0.035 0.035 0.035 0.035
dlnK 0.027 ** 0.026 * 0.027 ** 0.028 ** 0.028 ** 0.027 ** 0.014 0.014 0.014 0.014 0.014 0.014
P*M -0.014 -0.001 -0.004 0.022 0.012 0.010
P -0.004 -0.006 -0.003 -0.004 0.008 0.009 -0.010 0.010
M 0.007 0.008 0.013 0.009 0.010 0.018
P*M*E 0.206 0.086 0.127 *** 0.168 0.059 0.046
P*E 0.087 ** 0.088 * 0.045 0.059 0.044 0.053 0.052 0.054
M*E 0.019 -0.007 -0.120 0.069 0.076 0.157
E 0.001 0.031 0.003 0.018 0.002 0.020 0.036 0.030 0.034 0.032 0.036 0.032
(P+M+P*M)*E=0 (P+P*M)*E=0 F-test 7.87 *** 8.90 *** R2 0.491 0.489 0.491 0.493 0.491 0.491 N 1321 1321 1321 1321 1321 1321 Standard errors are in parentheses. These are long-differenced specifications. Value added, capital, and labor are annualized changes 2004-2007. Marketing innovation (M), product innovation (P), and organizational innovation (O). All regressions include industry dummies, regional dummies, ownership dummies, and export intensity.
-
Table 3: Firm Level Productivity Functions. Product, Marketing, and Organizational Innovation. Change in ln(Value added 2004-2007) (annualized mean =.024)
1 2 3 4dlnL 0.837 *** 0.841 *** 0.839 *** 0.838 *** 0.035 0.035 0.035 0.035
dlnK 0.028 ** 0.029 ** 0.029 ** 0.029 ** 0.014 0.014 0.014 0.014
P*M -0.014 -0.001 -0.004 0.022 0.012 0.010
P -0.003 -0.005 0.010 0.010
M 0.014 0.018
O 0.000 -0.001 0.001 0.000 0.008 0.009 0.008 0.008
P*M*E 0.196 0.077 0.100 ** 0.163 0.061 0.049
P*E 0.022 0.036 0.059 0.059
M*E -0.119 0.153
O*E 0.100 ** 0.079 0.079 0.086 * 0.049 0.052 0.053 0.050
E -0.024 -0.018 -0.033 -0.027 0.048 0.040 0.047 0.048
F-test (P+M+P*M)*E=0 (P+P*M)*E=0 3.32 * 4.53 ** R2 0.491 0.495 0.493 0.493 N 1321 1321 1321 1321 Standard errors are in parentheses. These are long-differenced specifications. Value added, capital, and labor are annualized changes 2004-2007. Product innovation (P), marketing innovation (M), and organizational innovation (O). All regressions include industry dummies, regional dummies, ownership dummies, and export intensity..
-
Table 4: Firm Level Productivity Functions. Product, Marketing, and Organizational Innovation. Change in ln(Value added 2004-2007) (annualized mean =.024)
1 2 3 4dlnL 0.842 *** 0.840 *** 0.840 *** 0.830 *** 0.035 0.035 0.035 0.039 dlnK 0.028 ** 0.029 ** 0.040 *** 0.042 *** 0.014 0.014 0.014 0.016 PM -0.003 -0.001 0.003 -0.004 0.011 0.011 0.011 0.012 PNM -0.003 0.000 -0.006 -0.011 0.010 0.011 0.012 0.013 NPM 0.013 0.015 0.013 0.005 0.018 0.019 0.015 0.015 ONPNM 0.006 0.006 0.011 0.011 PM*E 0.131 *** 0.168 *** 0.155 *** 0.104 *** 0.047 0.052 0.048 0.037 PNM*E 0.045 0.082 0.098 * 0.048 0.052 0.057 0.057 0.052 NPM*E -0.120 -0.083 0.117 0.091 0.157 0.158 0.078 0.073 ONPNM*E 0.082 ** 0.085 * 0.043 0.050 0.049 0.045 E 0.018 -0.019 -0.025 0.032 0.039 0.039 F-test PM*E>O*E PM*E>O*E 2.53 * 1.99 * R2 0.493 0.494 0.507 0.462 N 1321 1321 1158 776 Standard errors are in parentheses. These are long-differenced specifications. Value added, capital, and labour are annualized changes 2004-2007. Firms performing: product-marketing innovation (PM), product innovation but no marketing innovation (PNM), marketing innovation but no product innovation (NPM), organizational innovation but no product or marketing innovation (ONPNM). All regressions include industry dummies, regional dummies, ownership dummies, and export intensity.
-
Table 5: Firm Level Productivity Functions. Product, Marketing, and Organizational Innovation. Change in ln(Value added 2004-2007) (annualized mean =.024)
1 2 3 4 5 base median weight IV
dlnL 0.840 *** 0.887 *** 0.816 *** 0.828 *** 0.821 *** 0.035 0.051 0.054 0.040 0.041 dlnK 0.029 ** 0.034 ** 0.006 0.036 ** 0.030 ** 0.014 0.015 0.032 0.015 0.014 PM -0.001 -0.008 0.008 -0.002 -0.002 0.011 0.019 0.014 0.014 0.012 PNM 0.000 0.008 -0.003 -0.005 0.000 0.011 0.016 0.015 0.011 0.011 MNP 0.015 -0.015 0.062 0.019 0.014 0.019 0.023 0.024 ** 0.016 0.019 ONPNM 0.006 0.003 0.015 0.006 0.005 0.011 0.015 0.014 0.012 0.011 PM*E 0.168 *** 0.248 *** 0.083 * 0.129 ** 0.183 *** 0.052 0.075 0.050 0.052 0.055 PNM*E 0.082 0.079 0.069 0.029 0.088 0.057 0.076 0.062 0.051 0.056 NPM*E -0.083 -0.141 0.208 * 0.079 -0.070 0.158 0.193 0.111 0.069 0.153 ONPNM*E 0.082 ** 0.048 0.163 ** 0.086 0.088 * 0.050 0.060 0.076 0.071 0.053 E -0.019 -0.009 -0.034 -0.012 -0.027 0.039 0.050 0.044 0.038 0.039 F-test PM*E>O*E PM*E>O*E PM*E>O*E PM*E>O*E PM*E>O*E 2.53 6.88 *** N.A 0.35 2.60 * R2 0.494 0.477 0.533 0.511 0.472 N 1321 664 657 1321 1321 Standard errors are in parentheses. These are long-differenced specifications. Value added, capital, and labour are annualized changes 2004-2007. Firms performing: product-marketing innovation (PM), product innovation but no marketing innovation (PNM), marketing innovation but no product innovation (NPM), organizational innovation but no product-marketing innovation (ONPNM). All regressions include industry dummies, regional dummies, ownership dummies, and export intensity.
-
Appendix Table 1: Descriptive Statistics of sample Variable Obs Mean Std. Dev Min Max dln(value added) 1321 0.024074 0.189312 -1.21353 1.187493dln(value added) –dln(FTE) 1321 0.011242 ln(value added) 2001 1321 17.07981 1.57534 10.49904 22.89043ln(value added) 2004 1321 17.21692 1.516969 12.76903 22.92691product innovation 1321 0.39894 0 1product-marketing innovation 1321 0.124148 0 1product innovation, no marketing innovation 1321 0.274792 0 1organizational innovation only 1321 0.283876 0 1E2001 1321 0.166255 0.209093 0 1Export intensity 1321 0.260733 0.336472 0 1dln(capital) 1321 0.006381 0.379923 -1.95769 2.346803Dln(FTE) 1321 0.012832 0.152181 -1.25579 1.116635labor input (FTE) 1321 195.7033 513.3778 0 10047labor input (heads) 1321 194.0999 479.6713 1 9838Firms in primary sector 1321 0.003028 0 1Firms in manufacturing 1321 0.436033 0 1Firms in construction 1321 0.050719 0 1Firms in retail and wholesale trade 1321 0.245269 0 1Firms in transportation 1321 0.031794 0 1Firms in communication 1321 0.009841 0 1Firms in FIRE 1321 0.219531 0 1region1 1321 0.024981 0 1region2 1321 0.097653 0 1region3 1321 0.120363 0 1region4 1321 0.043906 0 1Ownership 1 1321 0.870553 0 1Ownership 2 1321 0.11355 0 1Ownership 3 1321 0.015897 0 1
battista forside.pdfJungeSevergniniSorensen_2012
top related